BBVA API Market
It sounds like science fiction, but the day has come when granting a loan at some credit institutions depends more on a customer's education or personality than on his or her financial rating. There are several startups today offering solutions that, thanks to algorithms, make it possible to reduce the levels of default of banks and lending companies or even to lend money with lower risk using other assessment models.
In our hyperconnected world, in addition to individuals and citizens, we are users. We leave a digital trail on the social media and other sources of information with our consumption habits, personal conduct… Some organizations that are in the business of lending money find in this data the assurance they need to grant a loan, even with more confidence than what capital or a stable job can offer.
Some of these startups are:
The letter of introduction of this lending company is compelling: “There are no complicated details hidden in the small print”. earnest uses technology and data to depart from the traditional loan granting model in the U.S: You can access loans under good conditions if your credit ratio is good. earnest operates differently: In addition to having a good financial rating, they consider aspects such as education, the employment history or the financial situation at the time of applying for a loan.
In the U.S., young newcomers to the labor market very likely lack a sound credit history, but their educational and professional projection is so good that it deserves trust in the granting of a loan. Thus, financial responsibility is more interesting than the credit history. earnest creates an individual profile of each applicant using algorithms and predictive analysis that assess both responsibility and potential. This is a software-based model.
The customers themselves participate in the creation of this profile: They provide information and data on their education, employment, income, debts, and financial and consumption habits. earnest checks this documentation thoroughly, granting loans to financially responsible customers. Any personal aspect in this granting decision is important.
Zestfinance was founded by Douglas Merril, who from 2003 to 2005 was Chief Information Officer and Vice President of Engineering at Google. His team is made up of former members of the giant search engine and Capital One, many of them mathematicians and computer engineers. Zestfinance uses all the experience acquired in the field of automatic learning at Google to apply the data to loans in an efficient way.
Zestfinance assesses many user variables to reduce the risk of fraud or default and establish a long-term business relationship with the customer. The company boasts that it has improved this ratio by 40%, which enables greater availability of credit, better interest rates and certainty of payment for the moneylenders or banking institutions which are its end customers. This is achieved through machine-learning algorithms and predictive analysis.
Today, most lenders use two techniques for risk assessment: logistic regression, which makes it possible to predict some variables based on others, and decision trees. According to Zestfinance, both models have limitations, for example, in the use of variables (only between 10 and 15 in the logistic regression model). This low use of variables results in an increase in the number of errors in the model.
Zestfinance offers a solution that combines automatic learning, management of large volumes of data and the experience gained by its team in credit management. Its decision-making model runs tens of loan applications in parallel with good results.
Affirm is the latest startup launched by Max Levchin, co-founder of PayPal, a convenient, long-term loan granting system for buying items such as household appliances or mattresses at retail stores without having to pay the interest associated, for example, with credit cards. Affirm imposes a 3, 6 or 12-month payment model.
In an article in Quartz, Levchin claimed that the idea behind Affirm is for customers to be able to buy at retail stores without having to use a card associated with hefty interest rates. This is in line with what younger users demand: Digital, modern forms of payment, far removed from the old systems such as credit cards.
What assessment system does Affirm use for granting loans? This startup has given up the most widely used methods such as FICO rating and uses social media like Facebook and LinkedIn to assess the customer's consumption and spending habits, financial responsibility, professional potential… This personal and public data helps the company create a more precise customer profile and reduce the risk of loan payment default.
Upstart also introduces changes to the traditional model based on FICO ratings. This startup uses this index, in addition to the credit history or professional income. But it also uses additional elements to create the customer's profile: University studies, educational training areas and analyses for preparing predictive models that assess the customer's willingness to pay.
Upstart's model simulates more than 50,000 possible scenarios for each customer, from which it assigns a given interest rate to each loan. If there are more possibilities of default, higher interest rates are imposed. In 2014 Upstart lent more than 100 million dollars.
Follow us on @BBVAAPIMarket
Online businesses are offering more and more facilities for customers to pay for their products. It is becoming more common to integrate financing into retail applications, and it is an added value for companies that implement it within their platforms.
Open banking is transforming the banking business as we know it, and the mortgage market was not going to be left out. Mortgages are the main business of many banks, which is why many of them are seeking to adapt their marketing to this new reality.